Date: Sat, 21 Jan 2012 00:22:18 +0000 From: Alex Sicamiotis <alekshs@...mail.com> To: <john-users@...ts.openwall.com> Subject: RE: DES with OpenMP > Date: Sat, 21 Jan 2012 02:15:24 +0400 > From: solar@...nwall.com > To: john-users@...ts.openwall.com > Subject: Re: [john-users] DES with OpenMP > > On Fri, Jan 20, 2012 at 09:35:26PM +0000, Alex Sicamiotis wrote: > > ...the OMP version is scoring +200k in many salts over the noOMP despite threads=1, > > Yes. > > > when it should be equal. > > No, it should not. It should be close, but not equal. Normally, OpenMP > build running with 1 thread is slightly slower than non-OpenMP build, > but it can also be the other way around. The code is different in many > ways. Even the chunk size (number of passwords being hashed per > crypt_all() call) differs by a factor of 32 between these builds. > Normally, this hurts performance when you don't actually run multiple > threads, but in certain builds and on certain machines it can be the > other way around as you have seen. > > > This implies a boost from the second core > > No, it does not. > > > (and top confirms this boost). > > It is tricky to actually confirm that with top. I think you're seeing > what you expect to see rather than what actually happens. > Top indicated a 3-5% spike when john started benchmarking (icc build) so I figured that these 200k must be it, assuming (erroneously) that the code is identical... Apparently it has significant differences that speed up the OMP version and the nonOMP version is somewhat suboptimal (?). I just turned off the second core (added maxcpus=0 to grub's menu) to check with only one core and indeed you are right. It's NOT performance from the second core. export OMP_NUM_THREADS=1 /john-GCC-O2-nocona-OMP -test Benchmarking: Traditional DES [128/128 BS SSE2-16]... DONE Many salts: 3749K c/s real, 3833K c/s virtual Only one salt: 3516K c/s real, 3595K c/s virtual ./john-GCC-O2-nocona-noOMP -test Benchmarking: Traditional DES [128/128 BS SSE2-16]... DONE Many salts: 3031K c/s real, 3112K c/s virtual Only one salt: 2944K c/s real, 3016K c/s virtual ./john-icc-OMP-fast-marchcore2 -test Benchmarking: Traditional DES [128/128 BS SSE2-16]... DONE Many salts: 4141K c/s real, 4234K c/s virtual Only one salt: 3733K c/s real, 3824K c/s virtual ./john-icc-noOMP-fast-marchcore2 -test Benchmarking: Traditional DES [128/128 BS SSE2-16]... DONE Many salts: 3902K c/s real, 3998K c/s virtual Only one salt: 3751K c/s real, 3843K c/s virtual
Powered by blists - more mailing lists
Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.